Skip to content
MatterSpace

MS Duplex — The Agent Operating System powered by Runtime Intelligence. Rent the model. Own the intelligence.

It learns from every confirmed result.

Confirm a result once, and the next request already benefits, with no retraining and no re-prompting.

Learning and personalisation. It runs After a confirmed result. In the Agent Operating System, it is the Runtime Intelligence that powers it all: learning from the work.

* Figures from MatterSpace’s internal benchmarks, not independently verified. Read the disclaimer

Confirm a fix once, and MS Duplex applies it from the next request.

You confirm a fix in one session: convert timestamps at the API boundary.

Step 1 of 4: A fix

It learns the moment you confirm.

The confirmation is the lesson, with no retraining run to wait for and no prompt to rewrite.

Step 1 of 4: Confirmed

You confirm a result the way you already do: you accept it, approve it or mark it right.

The next answer starts from the confirmed one.

Once MS Duplex has learned a result, the requests it fits no longer start from zero. Your AI is briefed with it, and when it is settled, MS Duplex answers them itself.

Step 1 of 3: First time

The first time, your AI works it out, and you confirm the fix.

Built to keep learning for months.

Learning from live work only helps if it holds up over time. MS Duplex met all five requirements on every one of ten runs.*

Step 1 of 6: New work

It learns new kinds of work as they arrive.*

See it on your own work.

From your own assistant to every agent your company runs, a pilot puts MS Duplex on your own work, with the measures agreed before it begins.